The demand for data storage and processing continues to grow, necessitating efficient, reliable, and cost-effective data centres. This data storage and processing demand, through such applications as AI, data mining and increased processing speeds is translating to increased power and water requirements for facilities, resulting in unprecedented growth and placing significant demands on already constrained power and water networks.

In this thought piece, we delve into the critical aspects of data centre management, focusing on tiering or rating vs. costs, Power Usage Effectiveness (PUE), Water Usage Effectiveness (WUE), and their relationship to data centre utilisation.

Tiering or Rating vs Capex/Opex

Data centre tiering or rating is defined by the Uptime Institute or the ANSI/TIA-942 Standard, respectively. These standards offer an industry benchmark for the classification of data centres, and their ability to maintain continuous operation, permit concurrent maintainability or where required offer fault tolerant operation.

The tiering or rating of a colocation data centre influences the type of customers it attracts, as these classifications align with the technical parameters specified within Service Level Agreements (SLA’s).

The tiering performance of each sub-system is typically informed by the customer SLA’s and generally, higher-tiered or rated data centres require greater capital expenditures (capex) for construction. This often results in additional maintenance capex (mainly replacement capex) and operating expenditure (opex) related to component inspections, given the increased equipment and components that must be monitored and maintained.

However, these initial investments may reduce opex over time due to reduced downtime and maintenance costs resulting from fewer reactive maintenance events, which can be costly.

PUE / WUE and Utilisation

Power Usage Effectiveness (PUE) is a key metric for assessing a data centre’s energy performance. It is calculated as an annualised value, dividing the total energy consumed by the data centre, by the energy consumed by the IT equipment. A lower PUE indicates higher energy efficiency. Advanced cooling technologies, efficient power distribution, and effective management practices can contribute to achieving a lower PUE.

Water Usage Effectiveness (WUE), equally measured as an annualised value, measures how effectively a data centre uses water for IT operations. This sustainability metric is particularly significant in regions where water availability for cooling is limited. WUE is calculated as the annual water usage divided by the energy consumed by the IT equipment. A lower WUE translates to more efficient water usage.

Methods to lower WUE include using air-cooled mechanical services systems in lieu of water cooled (though this may increase PUE) and, if using water cooling, implementing closed-loop water systems to minimise water loss through water evaporation or leakage.

A common theme relating to both PUE and WUE is data centre utilisation, which is linked to customer uptake and the use of leased white space. Data centres tend to exhibit higher operational PUE and WUE at the start of their operations due to lower utilisation compared to their design PUE and WUE. At this early stage of growth, the implementation of capital plant to match load growth is important in balancing energy performance and load growth. As utilisation increases, PUE and WUE will eventually reach an  optimum level, dictated by the operational efficiency curves of the installed components.

Conclusion

In conclusion, efficient data centre management hinges on understanding the intricate balance between tiering or rating, associated capex and opex costs, and the facility effectiveness metrics of PUE and WUE.

The tiering or rating determines the initial capital outlay and ongoing operational costs but can lead to long-term savings through enhanced reliability and reduced downtime. Meanwhile, achieving optimal PUE and WUE is crucial for sustainability and operational efficiency, particularly as data centre utilisation increases over time.